In all logistic regression models, variables used in the construction of the sampling weights (sex, age and residential location) were forced into the model to reduce biases in the other coefficients.
In that case, the model explains nominal variables using on average much smaller shocks.
Lipid, lipoprotein, anthropometry and age variables were included in the models as categorical variables, using tertiles based on participant distribution.
Intervention effects were analyzed using General Estimate Equation (GEE) modeling for count variables using the intention-to-treat principle.
The ten most frequently presented problems were modeled as dummy variables using cough as the reference category.
Following this, we introduce the analytical strategy employed, which is propensity score matching, and move on to describe the data, model and variables used.
In the final adjusted model all variables used for adjustment were also centered at their means.
In model 1, variables used in Figures 2 and 3 were used as dependent variables (age, high blood pressure, hypertensive drug use, smoking status, eGFR quartiles, UAER quartiles).
GAM is a flexible regression technique that models nonlinear relationships between variables using non-parametric smoothers.
Agrawal, A., Neale, M. C., Jacobson, K. C., Prescott, C. A. & Kendler, K. S. Illicit drug use and abuse/dependence: modeling of two-stage variables using the CCC approach.
